This is a general feature of procedural audio. In fact Pd plays very nice and has a fixed (predictable) cost for signal rate graphs - but a variable and unpredictable cost for message domain computation.
Proc audio eventually beats sample replay on cost because of dynamic level of detail where we get a variable (but predictable) cost for dynamically built signal graphs against a linear fixed cost for sample playback. It's an interesting bit of computer science to think about.
I know from talks with EA guys that EAPd ran into some problems and its performance was not spotless. But not for the reasons you state.
Andy
On Mon, 9 Jun 2008 20:55:58 +0200 Cyrille.Damez@laposte.net wrote:
On Monday 09 June 2008 06:23:04 Kyle Klipowicz wrote:
Maybe EA had better hire a genuine pro from the pd-list (wink wink).
I hope that this thread takes off, since I'm curious what others think on the topic. Spore could be the PR break that Pd has been waiting for!
One thing to note though is that they confessed having only used Pd to trigger midi events and play samples. There is no "actual sound synthesis" done by pd in Spore, only notes and messages generation. This was mostly for efficiency reasons, and because sound processing had to use only a predictable amount of cpu ressources. It is likely that any other game will have similar limitations.
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list